Transcribing

Bringing in the audio

Under Audio / SRT, either:

• drop an audio file onto the file field (its folder icon browses, too),
• click Browse audio…, or
• click Use attached audio to transcribe the sound already attached to the scene by VoxMark.

The line under the field shows which file will be transcribed. Switch to another scene and it follows that scene: its attached audio, or nothing until you pick a file. Dropping an .srt on the same field imports it instead — see SRT import & export.

Accepted formats: WAV, AIFF, MP3, MP4, M4A, MOV, OGG, FLAC and WebM. VoxMark sends the file to your provider as it is, so anything the provider accepts works. Only WAV gets the extra timing checks described below.

For the best transcript, use the voiceover on its own — music and effects make it worse.

The audio track

When you bring in a WAV, AIFF, MP3, MP4 or MOV, VoxMark attaches it to the scene so it plays with the timeline: it adds a null called VoxMark Audio at the top of the Object Manager with a Sound track pointing at your file. Bring in a different file later and the same track is pointed at the new one, rather than stacking another. M4A, OGG, FLAC and WebM are transcribed but not attached.

Options

Model (Groq only) — Whisper Large v3 Turbo, the default, is faster and places markers more accurately. Whisper Large v3 is slightly better word for word. A wrong word is a double-click to fix in the Navigate tab; a marker in the wrong place isn’t, which is why Turbo is the default.

Language — Auto-detect works for most recordings. Picking the language helps with short or mixed clips. The fourteen most common languages are at the top; all 99 that Whisper knows are below the divider.

Detail — Segments gives one marker per phrase; Words gives one marker per word. If a model returns no word timings, you get one marker per phrase and the log says so.

Max chars — splits captions longer than this at word boundaries, sharing the time out by word count (0 = no limit; the default is 60). A number stays with the word after it, so you get “over / 20 hours”, not “over 20 / hours”. It applies to SRT imports too.

Clear all existing markers first — removes every marker in the scene, including ones you placed yourself, before the new ones land. Leave it off and VoxMark replaces only its own markers; yours are left alone.

Transcript only (no markers) — transcribes and saves the transcript without touching your timeline. Add markers later with Create markers from transcript.

VoxMark remembers these options between sessions.

Speakers

Identify speakers (AssemblyAI) is on by default and only available with AssemblyAI — with another service the box shows unticked and greyed, and your choice returns when you switch back. Each voice gets its own name and marker colour. A recording with a single speaker is left unlabelled.

If you know how many people are talking, enter it in Speakers — an exact number (3) or a range (2-4), up to 20. Leave it blank for auto. Left on auto, AssemblyAI can merge two similar voices into one; a wrong exact number makes it worse, so give a range if you’re unsure.

Groq and OpenAI can’t tell speakers apart — their captions come in with no voice — but you can assign voices by hand with any provider — see The Navigate tab.

Transcribe

Click Transcribe. The button reads Transcribing… while it works, and the line under it shows each step. Cinema 4D stays usable while you wait. Only one transcription runs at a time.

When it finishes, VoxMark saves the transcript in the scene, lays the markers (unless you chose transcript only) and opens the Navigate tab. Transcribing again replaces the saved transcript — the old one is kept first; see Where VoxMark keeps things.

The two buttons under Transcribe:

• Rebuild transcript from markers — reads the VoxMark markers on the timeline back into a transcript, replacing the saved one.
• Create markers from transcript — lays markers from the saved transcript.

Both are explained on Markers in Cinema 4D.

The log

The folded Log at the bottom of the Transcribe tab shows every step of the last transcription or import — the file size, any silence trimmed, what was sent, and how many captions came back. Copy log puts it on the clipboard, ready to paste into a support ticket.

How accurate is the timing?

Whisper tends to report a caption’s start early wherever there’s a pause, sometimes by half a second. VoxMark corrects every caption against the word timings the transcription returns and, for a WAV transcribed in Segments detail, against the audio itself — measuring where speech actually begins. Where the audio can’t tell, the transcription’s own timing is kept; VoxMark never guesses.

A WAV that starts with a second or more of silence has the silence trimmed before upload, which avoids a known Whisper problem with long silent openings. The trimmed time is added back, so the markers still land where the words are in your file.